混合logit模型及其在合作医疗保险模式选择中的应用

摘要

Objective To apply mixed logit model for analyzing the data of new rural cooperative medical with suitability and identify the factors affecting the residents' choices of insurance mode. Methods Hypothesis test of IIA was performed using the mogtest module of StatalO.O to test the eligibility of the condition. The mixed logit model was established to allow the parameters to vary in the population using SAS9.1 MDC module. Results The data in this study did not satisfy the HA assumption (P＜0.01), so that the multinomial logit model was not applicable. The adjusted Estrella of the mixed logit model was 0.6658. Conclusion The mixed logit approach does not rely on the restrictive IIA assumption and allows for correlation patterns between choices and individual variation. This approach can help in the determination of the choices in new rural cooperative medical system.%目的 针对独立不相关(ILA)假设的限制,探讨混合logit模型在新型农村合作医疗中的适用性,并分析居民新型农村合作医疗保险模式选择的影响因素.方法 从选择项服从IIA假设人手,分析多项logit模型是否符合IIA假设条件.调用stata10.0软件的mlogtest模块进行IIA假设检验.再调用SAS9.1软件的MDC模块,建立了随机参数的混合logit模型.结果 在本次研究中,资料不能满足IIA假设(P＜0.01),不能应用多项logit模型进行分析;建立的混合logit模型的调整Estrella值为0.6658.结论 混合logit方法完全摆脱了ILA假设的限制,允许选项之间存在相关性,同时可以体现不同居民保险模式选择的异质性,该方法在新型农村合作医疗中具有一定的适用性.